Skip to contentSkip to conversation

sha256/gcrypt fixes

PATCH In review Eric Wong 2 participants 5 messages Jul 31, 2023

lore
Share & tools
lore

Eric Wong

author
#0 lore raw

cover sha256/gcrypt fixes

I noticed problems requiring patches 2 and 3 while eyeballing the code, but had to come up with the first one to fix SANITIZE=leak, first.

Eric Wong (3):
sha256/gcrypt: fix build with SANITIZE=leak sha256/gcrypt: fix memory leak with SHA-256 repos sha256/gcrypt: die on gcry_md_open failures

 sha256/gcrypt.h | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)
author
#1 lore raw

PATCH 1/3 sha256/gcrypt: fix build with SANITIZE=leak

Non-static functions cause undefined reference' errors when building with SANITIZE=leak' due to the lack of prototypes. Mark all these functions as `static inline' as we do in sha256/nettle.h to avoid the need to maintain prototypes.

---
 sha256/gcrypt.h | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
sha256/gcrypt.h +4 -4
index 501da5ed91..68cf6b6a54 100644
@@ -7,22 +7,22 @@
 
 typedef gcry_md_hd_t gcrypt_SHA256_CTX;
 
-inline void gcrypt_SHA256_Init(gcrypt_SHA256_CTX *ctx)
+static inline void gcrypt_SHA256_Init(gcrypt_SHA256_CTX *ctx)
 {
 	gcry_md_open(ctx, GCRY_MD_SHA256, 0);
 }
 
-inline void gcrypt_SHA256_Update(gcrypt_SHA256_CTX *ctx, const void *data, size_t len)
+static inline void gcrypt_SHA256_Update(gcrypt_SHA256_CTX *ctx, const void *data, size_t len)
 {
 	gcry_md_write(*ctx, data, len);
 }
 
-inline void gcrypt_SHA256_Final(unsigned char *digest, gcrypt_SHA256_CTX *ctx)
+static inline void gcrypt_SHA256_Final(unsigned char *digest, gcrypt_SHA256_CTX *ctx)
 {
 	memcpy(digest, gcry_md_read(*ctx, GCRY_MD_SHA256), SHA256_DIGEST_SIZE);
 }
 
-inline void gcrypt_SHA256_Clone(gcrypt_SHA256_CTX *dst, const gcrypt_SHA256_CTX *src)
+static inline void gcrypt_SHA256_Clone(gcrypt_SHA256_CTX *dst, const gcrypt_SHA256_CTX *src)
 {
 	gcry_md_copy(dst, *src);
 }
author
#2 lore raw

PATCH 2/3 sha256/gcrypt: fix memory leak with SHA-256 repos

In reply to Eric Wong

gcry_md_open' needs to be paired with gcry_md_close' to ensure resources are released. Since our internal APIs don't have separate close/release callbacks, sticking it into the finalization callback seems appropriate.

Building with SANITIZE=leak and running `git fsck' on a SHA-256 repository no longer reports leaks.

---
 sha256/gcrypt.h | 1 +
 1 file changed, 1 insertion(+)
sha256/gcrypt.h +1 -0
index 68cf6b6a54..1d06a778af 100644
@@ -20,6 +20,7 @@ static inline void gcrypt_SHA256_Update(gcrypt_SHA256_CTX *ctx, const void *data
 static inline void gcrypt_SHA256_Final(unsigned char *digest, gcrypt_SHA256_CTX *ctx)
 {
 	memcpy(digest, gcry_md_read(*ctx, GCRY_MD_SHA256), SHA256_DIGEST_SIZE);
+	gcry_md_close(*ctx);
 }
 
 static inline void gcrypt_SHA256_Clone(gcrypt_SHA256_CTX *dst, const gcrypt_SHA256_CTX *src)
author
#3 lore raw

PATCH 3/3 sha256/gcrypt: die on gcry_md_open failures

In reply to Eric Wong

`gcry_md_open' allocates memory and must (like all allocation functions) be checked for failure.

---
 sha256/gcrypt.h | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
sha256/gcrypt.h +3 -1
index 1d06a778af..17a90f1052 100644
@@ -9,7 +9,9 @@ typedef gcry_md_hd_t gcrypt_SHA256_CTX;
 
 static inline void gcrypt_SHA256_Init(gcrypt_SHA256_CTX *ctx)
 {
-	gcry_md_open(ctx, GCRY_MD_SHA256, 0);
+	gcry_error_t err = gcry_md_open(ctx, GCRY_MD_SHA256, 0);
+	if (err)
+		die("gcry_md_open: %s", gcry_strerror(err));
 }
 
 static inline void gcrypt_SHA256_Update(gcrypt_SHA256_CTX *ctx, const void *data, size_t len)
#4 lore raw
In reply to Eric Wong
Quoted 3 lines · Eric Wong

I noticed problems requiring patches 2 and 3 while eyeballing the code, but had to come up with the first one to fix SANITIZE=leak, first.

Thanks.

Quoted 8 lines

Eric Wong (3):
sha256/gcrypt: fix build with SANITIZE=leak sha256/gcrypt: fix memory leak with SHA-256 repos sha256/gcrypt: die on gcry_md_open failures

 sha256/gcrypt.h | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)